The invention relates generally to networks with application computers or servers, and more particularly to automatic programming of the application computers or servers to perform corresponding services to users via the network.
An IP network such as the Internet network conforms to a TCP/IP protocol suite for interconnecting a number of other IP networks. TCP is a transport layer protocol providing end-to-end data transfer. IP is an internetwork layer protocol hiding the physical network architecture to form what appears to the user to be only one large network. In an IP network, two computers can either be directly or indirectly connected together. A physical link can provide a direct connection between a pair of computers. Two or more computers can be connected through a switch on a single physical cable. The physical cable can be connected to a switch port, such that the computers are associated with the switch port.
A Virtual Local Area Network (VLAN) is known. A VLAN can be formed from a multiport switch having a multiplicity of internal switch elements such as transistors. By controlling the internal switches, the external computers connected to the switch can be interconnected. The external computers which are all interconnected by the switch elements to communicate with each other are considered to reside on the same VLAN. In such an implementation, a management computer controls the virtual connection (or disconnection) of several computer using switch configuration commands and without manual intervention. The switch is configured in such a way that only selected switch ports are connected together.
Even though it is possible to build VLANs including computers which can be changed over time, each of the computers has a specific function within the network. If the workload increases, the computers can be reconfigured, but this may be a long and tedious process requiring manual intervention. This limits the adaptability of the network when the demand varies.
Some existing technologies such as grid computing share processing over a pool of computers. The workload is divided into separate units which can be processed independently by separate computers. Each computer sends back the result of its computation when it is finished. The sharing of the physical resources and its split between the different computers that participate to the grid is seamless to the user. Should the user's need for computing power increase, the system will adapt by sending work to more computers. Conversely, when the need for processing power decreases, the computers can be used for other purposes. The system is therefore flexible and adaptive because it is able to automatically adjust to users' needs. However, processing power is only a part of the problem that the users face. Indeed, the system's adaptability is limited by the established framework of the network. Thus, when a new processor is assigned to some task to share the load of an overloaded processor, the new processor is added to the existing one, and the network keeps the same basic structure. However, the need for adaptation may also involve the rework of the network structure in terms of the functions that perform the various network elements.
An object of the present invention is to facilitate the change of function provided by computers on a network.